In the Tasmanian rock lobster assessment model, as in many other assessment models, selectivity is treated as a fixed effect. The actual size composition in the population is estimated by dividing the survey catch in each size class by the size-specific selectivity. Recently, it has been found that larger lobsters tend to inhibit smaller lobsters from entering traps. We suggest that, as larger lobsters are removed from the population by harvesting, the number of smaller lobsters in the catch increases because they become more catchable. To determine the effect of a change in selectivity pattern on our perception of population composition, we applied selectivity curves that accounted for the effects of a change in size composition to sampling data from a population of lobsters that had seen a substantial decline in large legal-sized lobsters due to harvesting over the last 35 years. The results suggest that recent recruitment is lower than the recruitment that occurred in the 1960s, but this is masked in the unadjusted sampling data by changes in selectivity. This could contribute to masking of a stock–recruitment relationship. If this is a common phenomenon in crustacean trap fisheries, the assessment approaches now in use may be more risk prone than previously realised.